Reacti ting ng flow modeling ng and applica cati tions ns in - - PowerPoint PPT Presentation

Reacti ting ng flow modeling ng and applica cati tions ns in STAR-CCM+

Yongzhe gzhe Zhang, ng, CD-adapco pco

Bet etter er flow and mixing ng accurac racy Results lts in bett etter er predic iction ion with th PVM combus bustion tion model dels ~32.7 .7 million ion cells ls Δt = 1x10-6 s

LES: : Scaled d Combust ustor

LES Flare: : Impr mproved d predi dicti ction n of combus usti tion n effici ciency ncy

PVM model el ~15 million lion cells ls Δt = 5x10-5 5 s

A Validation of Flare Combustion Efficiency Predictions from Large Eddy

• Simulations. Anchal Jatale, Philip J.

Smith, Jeremy N. Thornock, Sean T Smith, Michal Hradisky. University of

• Utah. Combustion and Flame.

More Large ge Eddy y Simulatio mulation n (LES) ES)

– Better prediction of instantaneous flow characteristics and turbulence structures – Computationally expensive

Include ude Detailed d Chemi mistr try

– Better prediction of autoignition and emissions (CO/NOx)

– Models

• Complex Chemistry model
• Tabulated Chemistry model

Applicat ication ion trend

Complex Chemistry Model

Dars-CFD FD Chemistry mistry reducti ction : Offli line (DRG) Turbule lence ce- chemi mistry stry Interactio ction Effici icient t ODE solve ver Computa tati tional Cost Storage/Re /Retrie trieva val l Sch cheme me(IS ISAT AT) Eddy y Dissip ipati tion Conce cept (EDC) Load bala lancin cing for parall llel l compu mputi ting Analyti lytica cal l Jacob cobia ian Equil ilib ibriu ium m Time Scale le (Init itia iali liza zati tion)

Transport equation of chemical species Nonlinear, stiff ordinary differential equations (ODEs)

Equi quili libri brium um Time e Scale e Model (EqTS TSM)‏

Motiva vati tion

– A better initial condition can greatly accelerate DARS-CFD

Model

– The model assumes the species composition to relax towards the local chemical equilibrium at a characteristic time scale determined based on the local flow and chemistry time scales – Quickly provides an reasonable initial condition to DARS-CFD – Results similar to PPDF equilibrium, but more flexible:

• no stream limitation/no precomputed table needed/easier to set up

– Can be used as a standalone model to obtain a quick approximate solution

Motiva vati tion

– Detailed chemistry is important to predict autoignition and emissions (CO/NOx) – Computationally expensive to include a full set of species

Tabulated d Detailed d Chemi mistr stry y for turb rbul ulent nt combustion ustion

– Precompute chemistry table and retrieve during CFD computation

• Can use large mechanism

– Dimension reduction to chemistry – Consider turbulence-chemistry interactions.

Existi ting ng models

– PPDF with equilibrium – PPDF with laminar flamelets – PVM (Progress variable model) – FGM (Flamelet Generated Manifold)

Tabula ulated ed Chemis mistr try y Model

Simi milar r to the existi sting ng PVM model:

– A tabulated detailed chemistry model – A progress variable is used to bridge the CFD side and the table

Impr mproveme ment nts s comp mpared d to the exist sting ng PVM model

– Table is from flamelet manifold

• A turbulent flame is an ensemble of laminar flamelets

– Option of using progress variable variance

• Presumed Beta PDF in progress variable space

– Option of considering heat loss ratio – Flexible progress variable definition

• Chemical enthalpy

– Sum over all species

• Species weights

– Defaults: YCO+YCO2

FGM combus bustion tion model

FGM table le genera eratio tion in DARS-BASI ASIC

9

• Generated table can directly be loaded into STAR-CCM+ for

further construction

Validation tion with h IFRF F glass s furnace nace

Heat loss effect is important

11

Latest t model l additions tions (v 9.04 04-10.0 0.04) 4)

Includ lude e det etailed ed chemist mistry y with an affor

• rdabl

able e computation putational al cost

– Equilibrium Time Scale – Flamelet Generated Manifold (FGM)

Cope e with h more e comple plex conf nfigurat igurations ions

– Inert stream – Reacting channels

Expand nd appli licati cation

• n coverages

erages

– Polymerization – Surface chemistry with multiple sites and open sites

Reacti cting Channel

Reacti tion n models s in STAR-CCM+

Reactio ction Models ls Multi ti-co compo mponent t Gas Lagrangia ian Multi tiphase se Eule leria ian Multi tiphase se Multi ti-co compo mponent t Liquid id Non-Premixe Premixed Combust stio ion Premixe mixed Combusti stion Parti tiall lly-Pre Premixe mixed Combust stio ion Emissio ssion Models s (Soot/ t/NO NOx/CO) x/CO) Eddy y Conta tact ct Model l (ECM) Polyme ymeriza izati tion Parti ticle cle Reactio ction Coal l combu mbusti stion Interphase se Reactio ction Surfa face ce Chemistry mistry

13

Latest t model l additions tions (v 9.04 04-10.0 0.04)

Includ lude e det etailed ed chemist mistry y with an affor

• rdabl

able e computation putational al cost

– Equilibrium Time Scale – Flamelet Generated Manifold (FGM)

Cope e with h more e comple plex conf nfigurat igurations ions

– Inert stream – Reacting channels

Expand nd appli licati cation

• n coverages

erages

– Polymerization – Surface chemistry with multiple sites and open sites

Inert t stream eam for r PPDF combu mbust stion ion model el

Motiva vati tion

– To reduce the PPDF table size for complex configurations where one stream, or part of the stream, is inert (negligible reactivity and sole effect is for dilution)

Inert t str tream m treatm tment nt

– Only consider its dilution effects to the reacting mixture – Compared to take it as active

• Smaller table size
• Faster table generation
• Faster interpolation

Inert t strea tream m model

– A transport equation for the mixture fraction solved for inert stream – Species mass fractions from reacting and inert streams – Temperature from local total enthalpy and mean species

Reacti cting ng Channel nel Co-Si Simu mulation ation

Applic ication ation

– Process heaters – Cracking furnaces – Steam reformers

Modeling eling Challen llenges es

– Firebox side has multiple burners – Process side has many tubes – Full 3-D modeling is computationally intensive

Perform

• rmanc

ance e Considerat ideration ions

– Uniform heat distribution – Emissions – Conversion rate

Modelin eling of Proces ess side

Comput mputationally expe pensi sive Comput mputationally less s expensi sive

3-D vs 1-D

Gas-Pha Phase: e: [ FireB eBox Side] e]

– 3-D, turbulent flow – Combustion models – Heat transfer

Reacti ting g Chann nnel: el: [Proc

• ces

ess Side] de]

– 1-D Plug Flow Reactor (PFR) – Inlet composition, temperature – Process-side reactions – No meshing, solving with STAR-CCM+

Couplin ling

– Temperature is provided to the process side – Heat flux is returned back to firebox side

Reacti cting ng Channel nel Co-Si Simu mulation ation

Burner er Proces ess Side Side An elega gant nt way to fully ly couple le Firebo rebox side e and Proces ess side

Output put from

• m Co-simul

simulation ation : Process ess Side

Axial l distr trib ibution ution of Temperature, erature, Heat t Flux, , and Species ies convers ersions ions

CH4 Mass Fraction H2 Mass Fraction

Polymeri ymeriza zation tion

Expand and our applica cati tion n coverage age Polym ymeriza rizati tion

• n Process

cess

– monomers are linked by chemical reactions to form long chains – starts with mixing a Monomer (M) and an Initiator (I) in a Solvent (S) – Steps involved: initiation/propagation/transfer/branching/termination – Final product is polymers of varying lengths and structure.

Polym ymerizati zation

• n Moment

nt Model for free radical cal polym ymerizati tion

– Scalar Transport Equations for Moments are solved in STAR-CCM+: live/dead polymers – source terms of the above moment transport equations depend on the sub processes of polymerization. – Provide: total polymer concentrations, NACL/NAMW, WACL/WAMW, polydispersity index

Industri ustrial-Sca Scale Stirred d Tank k Reactor – Styrene ne Polym ymerization zation

• Steady (Implicit Unsteady)
• K-Epsilon Turbulence
• Realizable K-Epsilon Two-Layer
• Two -Layer All Y+ Wall Treatment
• Multi-Component Liquid
• Polymerization
• Segregated Flow
• Segregated Fluid Enthalpy
• Three Dimensional
• MRF, RBM

Polyd ydisp ispersit sity index

Multi tiple ple sites s for surface ce chemi mistr try Chemical vapor deposition (CVD) reactor

Open sites es for r surface ce chemi emistr stry

Adsorption ption reacti tion n descr cripti tion

– Atomic Site

• AsH3(g)+Ga(s)->AsH3(s)+Ga(b)

– Open Site

• O(s)+AsH3(g)->AsH3(s)

Open sites s treatm tment nt

– Considered as a species – Contains no element (empty) – Named as OPEN in the CHEMKIN kinetics input file

Application ication extens nsions ions

Large ge Eddy Simul mulation ation (LES) S) with h det etailed ed chemist mistry

– Gas turbine combustors – Burners, Furnaces and Incinerators – Fires

High h speed ed flows ws

– Scramjet – Rocket engine nozzles

Multipha phase se react ctions ions

– Coal reactors: Pulverized/Fluidized bed – Surface chemistry (SCR/CVD)

Optimiz mizations tions

– Chemistry – Combustor design

Application ication extens nsions ions

Large ge Eddy Simul mulation ation (LES) S) with h det etailed ed chemist mistry

– Gas turbine combustors – Burners, Furnaces and Incinerators – Fires

High h speed ed flows ws

– Scramjet – Rocket engine nozzles

Multipha phase se react ctions ions

– Coal reactors: Pulverized/Fluidized bed – Surface chemistry (SCR/CVD)

Optimiz mizations tions

– Chemistry – Combustor design

Supersonic Combustion

• H2 Fueled

led NASA SA SCHOLA OLA direct rect-conn

• nnect

ect Scramje amjet t engine ne

• Valid

idate against experime iment nt and NASA SA VULCA CAN N code de Mesh: h: 1.4M Hex-dominant 10 Prism Layers Solver: er: Density based solver Steady,k-w SST Non-adiabatic PPDF

Supersonic Combustion (2)

Yongzhe Zhang, Ivana Veljkovic, Nolan Halliday and Rajesh Rawat, "Numerical Simulation of a Scramjet Using a Storage/Retrieval Chemistry Scheme", AIAA 2014, Washington DC

Coal Combust ustion n Valida dati tion

Model el Select ection ion:

– Coal particles

• Moisture evaporation
• Raw coal de-volatilization
• Char oxidation
• Fuel NOx + Thermal Nox
• Particle radiation

– Gas Phase

• 4 step global kinetics
• Radiation

– Participating Media Weighted Sum of Gray Gases Quantity Measured Predicted Dimension T 1353 1347 Kelvins Oxygen 3.0 3.08

• Vol. %, dry

CO2 15.6 15.43 Vol.%, dry Burnout 99.4 100.0 Weight %

Mathematical Modeling of a 2.4 MW Swirling Pulverized Coal Flame

• Combust. Sci. and Tech, 1997, Vol 122, pp. 131-182

Centerli rline ne Temp mperat rature ure

Applica cati tion n Example ple: : SCR modelling ng

• Lagrangian droplets are injected into the hot exhaust flow
• The liquid droplets and gas exhaust pass through a mixing vane
• Some of the droplets impinge on the vane and form a film which boils
• The mixture of exhaust gases and boiled vapour move into the catalyst

Opti timiza zati tion n of Gas Turbine ne using ng STAR-CCM+ M+ and Optimate+

• Match flame length and shape with experiments,
• Minimize NOx and CO emissions,
• Minimize pressure drop,
• Maximize combustion efficiency,
• Maximize homogeneity at combustor exit

Geometry Optimization or Operating Condition Optimization

Generic ric Combus ustor

• r for Optimizati

tion

Combus ustor

• r Type

e – Annular lar

– Optimize geometry based on performance objectives

Pa Paramet eteriz erized ed design ign features tures

– Swirler twist angle – Liner hole radius – Hollow cone injector’s

• Inner and outer cone angle

Pa Parame meter Ranges

Swirler geomet metry y for Min (16°), , Baseline (45°), , Max (93°) Hole radius s for liner: : Min=1mm, =1mm, Baseline=2mm, =2mm, Max=2.9 =2.9 mm Inner cone angle: : 0 to 45 degrees.

• s. Outer cone angle:

: 45 to 120 degrees

Opti timiza mizati tion n Resul sult

Case Baseline line (Rank k – 40) 40) Rank – 1

Twist st angle gle (°) 45 45 51 51 Liner er Hole Radius us (mm) 2 1 Inne ner Cone e Angle e (°) 10 10 37 37 Outer er Cone ne Angle e (°) 90 90 61 61 Cone ne Angle e (°) 80 80 24 24 Volume ume aver eraged ged T (K) 1000. 0.7 969.5 Tot

• tal CO (kg/s)

/s) 9.441E-07 07 8.613 13E-07 07 Tot

• tal NOx (kg/s

/s) 2.697E 7E-08 08 1. 1.126E 6E-08 08 Perfor

• rma

manc nce

• 2.00

00

• 1.

1.33 33

8.8% 58.2 .2%

New models ls added to (v v 9.04-10.04)

– Include detailed chemistry with an affordable computational cost

• Equilibrium Time Scale
• Flamelet Generated Manifold (FGM)

– Cope with more complex configurations

• Inert stream
• Reacting channels

– Expand application coverages

• Polymerization
• Surfaced chemistry with multiple sites and open sites

Appl plic icati ation

• n extensions

nsions

– Large Eddy Simulation (LES) with detailed chemistry – High speed flows – Multiphase reactions – Optimizations

Summa mmary

Thank nk you for r your r attent ention! ion!